major change, used biomed_llama 7b

app.py

@@ -191,6 +191,7 @@ def initialize_model():
 
 def initialize_llm():
     try:
+        print("Starting LLM initialization...")
         model = AutoModel.from_pretrained(
             "ContactDoctor/Bio-Medical-MultiModal-Llama-3-8B-V1",
             device_map="auto",
@@ -198,12 +199,13 @@ def initialize_llm():
             trust_remote_code=True,
             low_cpu_mem_usage=True
         )
+        print("Model loaded successfully")
         
         tokenizer = AutoTokenizer.from_pretrained(
             "ContactDoctor/Bio-Medical-MultiModal-Llama-3-8B-V1",
             trust_remote_code=True
         )
-        
+        print("Tokenizer loaded successfully")
         return model, tokenizer
     except Exception as e:
         print(f"Failed to initialize LLM: {str(e)}")
@@ -252,11 +254,13 @@ def process_image(image_path, text_prompts, modality):
         
         # Process with LLM only if available
         if llm_model is not None and llm_tokenizer is not None:
+            print("LLM model and tokenizer are available")
             try:
                 pil_image = Image.fromarray(image)
                 question = 'Give the modality, organ, analysis, abnormalities (if any), treatment (if abnormalities are present)?'
                 msgs = [{'role': 'user', 'content': [pil_image, question]}]
                 
+                print("Starting LLM inference...")
                 llm_response = ""
                 for new_text in llm_model.chat(
                     image=pil_image,
@@ -267,19 +271,22 @@ def process_image(image_path, text_prompts, modality):
                     stream=True
                 ):
                     llm_response += new_text
+                print(f"LLM generated response: {llm_response}")
+                
+                # Make the combined analysis more visible
+                combined_analysis = "\n\n" + "="*50 + "\n"
+                combined_analysis += "BiomedParse Analysis:\n"
+                combined_analysis += "\n".join(analysis_results)
+                combined_analysis += "\n\n" + "="*50 + "\n"
+                combined_analysis += "LLM Analysis:\n"
+                combined_analysis += llm_response
+                combined_analysis += "\n" + "="*50
                 
-                # Combine both analyses
-                combined_analysis = "\n\n".join([
-                    "BiomedParse Analysis:",
-                    "\n".join(analysis_results),
-                    "\nLLM Analysis:",
-                    llm_response
-                ])
             except Exception as e:
-                print(f"LLM analysis failed: {str(e)}")
+                print(f"LLM analysis failed with error: {str(e)}")
                 combined_analysis = "\n".join(analysis_results)
         else:
-            # If LLM is not available, only show BiomedParse results
+            print("LLM model or tokenizer is not available")
             combined_analysis = "\n".join(analysis_results)
         
         return results, combined_analysis